AI Incident Declaration & Stakeholder Mapping Intake
Week 1 intake form for the shared-responsibility team to define the incident taxonomy, declaration process, role assignments, and stakeholder map. Completed during Day 2-7 of the 30-day AI IR build and reviewed by the executive sponsor.
Part of: The 30-Day AI Incident Response Build: From Zero to Usable IR Capability with a Small Team
- Format
- FORM
- Fields
- 19
- Access
- Open
Form · 19 fields
AI Incident Declaration & Stakeholder Mapping Intake
Week 1 intake form for the shared-responsibility team to define the incident taxonomy, declaration process, role assignments, and stakeholder map. Completed during Day 2-7 of the 30-day AI IR build and reviewed by the executive sponsor.
Form ID
DOC-0009
Category
General
Access tier
FREE
Date
12 August 2026
Purpose
This intake captures the three deliverables your shared-responsibility team needs by end of Week 1: a working incident taxonomy that distinguishes AI incidents from general software incidents, a declaration process with named authority and triggers, and a stakeholder map with notification tiers. Fill it in once, review with your executive sponsor, and use the outputs directly in your escalation matrix (Week 2) and containment runbook (Week 4).
Incident Taxonomy
Define what counts as an AI incident in your environment so the team can distinguish it from a general software outage. This boundary determines when the AI-specific layer of your minimal general IR skeleton engages.
This determines whose harm you are designing for and directly shapes the severity tiers you will build in Week 2.
These are the failure modes your IR program must be able to detect and contain. Select all that apply to your production AI systems today.
Write a working boundary statement your team can apply under pressure. Example: 'An incident is AI-specific when the primary failure surface is model output quality or behavior — not application uptime or code defects — and containment requires model-level actions such as rollback, guardrail activation, or circuit-breaking rather than a code deploy.'
Declaration Process
Establish who can declare an incident, what conditions trigger declaration, and where coordination happens. This is the entry point to your entire IR skeleton — without it, detection signals from Week 3 have nowhere to route.
Name the individual(s) with declaration authority. With a 2-3 person shared-responsibility team, this is typically the primary on-call or the team lead. Use a person's name, not a role title that does not exist yet.
Select all conditions that should result in a declaration without further deliberation. Anything not listed here defaults to monitoring, not declaration — be deliberate about what makes the cut.
The dedicated Slack or Teams channel where declaration happens and all response coordination takes place. This channel should already exist per the playbook's prerequisites.
Shared-Responsibility Team
Assign IR functions to each of your 2-3 team members. With no existing IR roles to map to, each person will hold multiple functions. Three core functions need coverage during an active incident: Incident Commander (coordinates response, owns escalation decisions), Technical Responder (executes containment — model rollback, guardrail activation, circuit-breaking), and Communications Lead (stakeholder notifications, status updates).
Format: Name — primary function, secondary function. With a 2-person team, one person will hold two of the three core functions.
Format: Name — primary function, secondary function.
Complete only if your team has a third member. Format: Name — function(s).
Who is first to respond and holds declaration authority during their rotation. This person is the default Incident Commander unless explicitly handed off.
Who backs up the primary on-call and assumes the role if the primary is unreachable. Must be a different person from the primary.
Stakeholder Map
Identify every internal stakeholder who must be informed during an AI incident, along with their department, preferred contact method, and notification tier. This map feeds directly into your escalation matrix in Week 2 — gaps here become blind spots during a live incident.
Select all departments that must be represented. Use this as a coverage checklist before listing individuals — if a department is unchecked, confirm no one in that function needs to know before proceeding.
One stakeholder per line, formatted as: Name | Department | Contact Method (Slack/Teams/Email/Phone) | Notification Tier (1=immediate notification, 2=within 1 hour, 3=summary only). Tiers are provisional — you will align them to your severity classification framework in Week 2.
Get this document
Free to read here. Sign in to take a copy with you — it takes a minute.
Sign in to downloadDocument info
- Format
- FORMPDF export
- Access tier
- Free
- Category
- General
- Published
- 12 August 2026